Ann: Results from Data Confirmation Study, page-95

I think you are misunderstanding what I meant and what this Confirmatory Data Analysis was.

They don't do the entire study again and I didn't mean that they tested it on the additional 1300 coughs. My understanding is that they take the data from the first study and have an independent statistician analyse that dataset to confirm the findings. They don't re-run the entire study. Statistics are complex and there are many different statistical models that can be applied to a given dataset... and depending on the model, can often result in a different outcome.

When they say they in the statement they have an additional 1300 cough recordings they don't mean that they ran the study again on those coughs. It means they will be inserted into the learning algorithm to allow it to train itself.

In short, when you say "an extra 1300 coughs and instead of improving data it actually got worse" that is misleading - they were never meant to affect these results. Those 1300 additional cough recordings were not used and have no impact on the results of THIS confirmatory data analysis. They will help improve the algorithm going forward.